AQP1 is Involved in NLRP3-Related Microglial Polarization and Cognitive Impairment in Chronic Sleep Deprivation

Yanhong Xiong; Weidong Liang; Hong Zhu; Xilong Guan; Pengcheng Yi; Lieliang Zhang; Yueyang You; Yingchuan Hu; Xiuqin Rao; Jun Ying; Xifeng Wang; Fuzhou Hua

doi:10.1007/s10753-026-02498-3

. 2026 Apr 20;49(1):144. doi: 10.1007/s10753-026-02498-3

AQP1 is Involved in NLRP3-Related Microglial Polarization and Cognitive Impairment in Chronic Sleep Deprivation

Yanhong Xiong ^1,^2,^#, Weidong Liang ^3,^#, Hong Zhu ^1,², Xilong Guan ⁴, Pengcheng Yi ^1,², Lieliang Zhang ^1,², Yueyang You ^1,², Yingchuan Hu ^1,², Xiuqin Rao ^1,², Jun Ying ^1,², Xifeng Wang ^2,^5,^7,^✉, Fuzhou Hua ^1,^2,^6,^✉

PMCID: PMC13222902 PMID: 42008056

Abstract

Sleep deprivation is closely associated with neuroinflammation and cognitive dysfunction, yet its underlying molecular mechanisms remain incompletely understood. Aquaporin-1 (AQP1) has been implicated in microglial regulation, but its role in chronic sleep deprivation (CSD)–induced cognitive impairment remains unclear. A chronic sleep deprivation (CSD) mouse model and an in vitro BV2 microglial inflammation model were established. Learning and memory functions were assessed using the Y-maze, novel object recognition, and Morris water maze tests. Transcriptomic sequencing was performed to identify key regulatory genes. AQP1 expression, NLRP3 inflammasome–related responses, microglial polarization, hippocampal pathology, and neuroinflammatory responses were systematically evaluated using molecular, histological, and immunological approaches. In addition, high-throughput virtual screening and molecular dynamics simulations were employed to analyze the potential association between curcumin and AQP1. CSD induced significant impairments in learning and memory, accompanied by hippocampal neuronal injury and neuroinflammatory activation. Transcriptomic analysis revealed that AQP1 was markedly upregulated in the hippocampus of CSD mice and was predominantly localized in microglia. AQP1 knockdown significantly ameliorated CSD-induced cognitive deficits and neuronal damage was accompanied by reduced NLRP3 inflammasome–related readouts, and promoted microglial polarization toward the M2 phenotype, these effects were largely reversed by the NLRP3 agonist nigericin. Furthermore, curcumin attenuated microglial M1 polarization, reduced neuroinflammation, and improved cognitive performance, alongside changes in AQP1 expression and inflammasome-related measures. AQP1 is associated with CSD-related cognitive impairment and neuroinflammatory responses, accompanied by changes in microglial polarization and NLRP3 inflammasome–related signaling. Curcumin shows neuroprotective effects under CSD conditions; whether AQP1 is a direct molecular target requires further validation.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10753-026-02498-3.

Keywords: AQP1, Chronic sleep deprivation, Cognitive dysfunction, Microglial polarization, NLRP3 inflammasome

Introduction

Sleep is a fundamental physiological process that is essential for maintaining optimal cognitive function and overall health. However, with the increasing pressures accompanying rapid economic and social development, sleep deprivation has become a global public health issue and poses a significant threat to human health [1]. In a large-scale study in the United States, approximately 27% of college students reported poor sleep, and one-third reported insufficient sleep duration [2]. In China, 43.9% of college students sleep less than 7 h/day [3]. CSD is closely linked to the onset of various diseases, including type 2 diabetes, cardiovascular diseases, depression, and neurodegenerative disorders [4–6]. Moreover, insufficient sleep duration negatively affects cognitive ability, resulting in slower reaction times, impaired perception, and an increased risk of dementia [7–9]. These cognitive impairments involve multiple mechanisms, primarily reduced synaptic plasticity, oxidative stress, neuroinflammation, and gut–brain axis dysfunction [10–13]. Among these, neuroinflammation is considered one of the core pathological mechanisms underlying the cognitive dysfunction induced by sleep deprivation [12].

Neuroinflammation plays a critical role in the cognitive impairment induced by CSD, particularly through the activation and polarization of microglia in the central nervous system. Microglia are the primary resident immune cells in the brain, and upon stimulation by injury or inflammatory signals, microglia become rapidly activated and differentiate into two functional phenotypes: proinflammatory (M1 type) and anti-inflammatory (M2 type). M1 microglia exacerbate neuronal damage by releasing proinflammatory cytokines (such as IL-1β, TNF-α, and IL-6), whereas M2 microglia promote tissue repair and neurogenesis through the secretion of anti-inflammatory factors (such as IL-10 and Arg1) [14–16]. Therefore, microglial polarization, particularly the shift from the M1 to the M2 phenotype, is considered a potential therapeutic target to alleviate neuroinflammation and improve cognitive function. Studies have shown that the cognitive impairment induced by CSD is often accompanied by excessive activation of microglia and their polarization toward the M1 phenotype [1]. However, the specific molecular pathways through which microglial polarization is regulated in the context of CSD and how this leads to neuronal damage and cognitive dysfunction remain insufficiently explored.

Aquaporins (AQPs) are membrane channel proteins that regulate transcellular water transport and are essential for maintaining cellular homeostasis in the central nervous system [17]. Aquaporin-1 (AQP1), the first identified member of this family, is abundantly expressed in the choroid plexus and contributes to cerebrospinal fluid dynamics and brain water balance [18]. Increasing evidence suggests that AQP1 is also expressed in microglia and participates in neuroinflammatory regulation. Altered AQP1 expression has been associated with changes in microglial activation and polarization, which in turn influence neuronal injury and cognitive outcomes in neurological disorders [19, 20]. However, whether AQP1 contributes to microglia-mediated neuroinflammation and cognitive impairment under CSD remains largely unknown.

The NOD-like receptor family pyrin domain–containing 3 (NLRP3) inflammasome is a central innate immune signaling platform in microglia and has been implicated in neuroinflammation-associated cognitive dysfunction [21, 22]. Aberrant activation of the NLRP3 inflammasome promotes the release of proinflammatory cytokines and is associated with microglial polarization toward a proinflammatory phenotype, which may exacerbate neuronal damage [23–25]. Dysregulated NLRP3 signaling has been implicated in multiple neurodegenerative and neuroinflammatory conditions [26, 27]. Although both AQP1 and NLRP3 have been implicated in microglial inflammatory responses, their relationship in the context of chronic sleep deprivation (CSD) remains unclear. In particular, it is unknown whether changes in AQP1 are associated with NLRP3 inflammasome–related responses and microglial polarization under CSD conditions, and how these changes may contribute to neuroinflammation and cognitive decline. Addressing this question may help improve our understanding of the molecular basis of CSD-induced cognitive dysfunction.

Curcumin is a polyphenolic compound derived from the rhizomes of turmeric and has been widely reported to exhibit anti-inflammatory, antioxidant, and anti-apoptotic activities. Owing to its generally favorable safety profile in experimental settings, curcumin has attracted considerable interest in neurological disorders [28, 29]. Previous studies have shown that curcumin can improve learning and memory performance in Alzheimer’s disease models and has been associated with reduced apoptotic signaling [30], while in Parkinson’s disease models it has been linked to neurofunctional improvement, potentially involving neurotrophic pathways such as BDNF signaling [31]. Curcumin has also been reported to enhance synaptic plasticity–related markers (e.g., PSD95 and BDNF) and to be associated with modulate microglial polarization in certain injury models [29]. Additionally, curcumin has been reported to influence microglial polarization by modulating the microRNA-205-5p/KLF2/ATF2 signaling pathway, which has been suggested to help alleviate neuroinflammation in ischemic stroke models [32]. Despite accumulating evidence for its neuroprotective potential, how curcumin modulates neuroinflammation and cognitive impairment under chronic sleep deprivation (CSD) conditions remains insufficiently characterized.

In this study, a CSD mouse model and an in vitro model of lipopolysaccharide (LPS)-induced microglial inflammation were established to investigate the role of AQP1 in CSD-associated neuroinflammation and cognitive impairment, with particular attention to microglial polarization and NLRP3 inflammasome–related inflammatory responses. These findings suggest that altered AQP1 expression is associated with microglial polarization, hippocampal injury, and neuroinflammatory changes under CSD conditions. In addition, we explored curcumin as a pharmacological intervention in this context and evaluated its association with microglial polarization, neuroinflammation, and cognitive outcomes. Collectively, these results help clarify pathophysiological processes relevant to CSD-induced cognitive dysfunction and support further investigation of AQP1-associated inflammatory regulation as a potentially relevant therapeutic direction. Importantly, our data do not establish direct target engagement or binding specificity between curcumin and AQP1. Therefore, curcumin is discussed here as a pharmacological probe, and whether it directly binds to AQP1 will require dedicated target-engagement assays in future studies.

Materials and Methods

Reagents

Antibodies against GFAP (#3670), IBA1 (#17198), NeuN (#24307), and IL-1β (#63124) were obtained from Cell Signaling Technology (Danvers, MA, USA). Antibodies against AQP1 (ab168387), TNF-α (ab215188), NLRP3 (ab270449), CD86 (ab119857), and CD206 (ab64693) were purchased from Abcam (Cambridge, MA, USA). Antibodies against ASC (A24165) and cleaved Caspase-1 (A25308) were obtained from ABclonal Biotechnology Co., Ltd. (Wuhan, Hubei, China). Anti-GAPDH (60004-1-lg) and anti-IL-6 (HZ-1019) antibodies were obtained from Proteintech (Wuhan, Hubei, China). Lipopolysaccharide (LPS) was purchased from Sigma–Aldrich (St. Louis, MO, USA). Curcumin and the NLRP3 agonist nigericin (Nig) were obtained from Master of Small Molecules (MCE). Unless otherwise specified, all other reagents were of analytical grade.

Animals and Treatment

C57BL/6J male mice aged 6 to 8 weeks were obtained from Skobis Company (Henan, China). These animals underwent a 7-day adaptation phase inside a specific pathogen-free (SPF) facility prior to any testing. The animals were maintained on a 12 h alternating light/dark cycle at 25 ± 2 °C with 50 to 60% relative humidity. The mice were housed in polystyrene enclosures, and water was available ad libitum.

To minimize potential confounding effects related to sex-dependent hormonal variation, only male mice were used in this study. Mice were randomly assigned to the respective experimental groups prior to treatment. Behavioral assessments and related data analyses were performed by investigators blinded to group allocation to reduce observer bias. All methods were approved by the Ethics Committee Animal Board (Approval number: RYE2024032901) and complied with the organization’s protocols for experimental animal handling.

AQP1 protein expression was measured at four specific intervals after CSD (48 h, 24 h, 12 h, and 6 h). A total of 50 mice were randomly assigned to five groups, each of which contained 10 individuals: CSD-48 h, CSD-24 h, CSD-12 h, CSD-6 h, and a control group. Immunofluorescence staining of AQP1 was performed on three animals from each of the CSD-24 h and control groups to determine AQP1 localization.

Forty rodents subjected to CSD were randomly assigned to four cohorts of ten animals each to determine how decreased AQP1 expression alters cognitive performance, as follows: AAV-sh-AQP1 + CSD, AAV-sh-AQP1 + CSD (replicate), CSD, and a control. These cohorts were employed to determine how AQP1 silencing affects cognitive capacity.

Fifty mice were assigned to five additional cohorts, each containing 10 individuals, to examine the mechanistic contribution of AQP1 to the cognitive deficits induced by CSD. The cohorts included AAV-sh-AQP + Nig (NLRP3 agonist) + CSD, AAV-sh-AQP1 + CSD, AAV-sh-NC + CSD, CSD, and control groups. Following CSD modeling, the NLRP3 agonist (4 mg/kg, i.p.) was delivered to the Nig group of animals [33], allowing evaluation of the effects of the NLRP3 pathway on cognitive dysfunction.

Additionally, 30 mice were randomly assigned to three cohorts, namely, Curcumin + CSD, CSD, and control, to examine how curcumin improves cognition through AQP1. Commencing with the initial week of CSD onset, 50 mg/kg curcumin was intraperitoneally administered each day for two weeks, concluding alongside the CSD procedure [34]. After CSD exposure, all animals were subjected to behavioral evaluations. All experimental protocols prioritized limiting both animal distress and the number of subjects involved.

Chronic Sleep Deprivation Protocol

An automated apparatus sourced from Shanghai Xin Ruan Information Technology Co., Ltd. (Shanghai, China) facilitated persistent sleep restriction in rodents via protocols established previously [8, 35–37]. Randomly timed bar rotations within the device blocked animal repose. Five rodents per cohort were placed in this device and provided food and water ad libitum. The animals were acclimatized to the apparatus 7 days prior to testing in accordance with documented procedures. Bar spinning was employed at 5 revolutions per minute from 14:00 to 10:00 the following day, lasting for a total of 20 h. This was followed by a 4-h rest period that started at 10:00. Cohorts designated as controls utilized matching hardware but were allowed routine rest aligned with designated intervals. The experiment lasted 21 days [38]. Because EEG/EMG recordings were not collected, sleep architecture (e.g., NREM/REM staging) was not directly quantified in this study; the CSD paradigm was implemented according to previously validated protocols [8, 35–37].

Intracerebroventricular (ICV) Injection

The production of adeno-associated viruses (AAVs) containing siRNA1 targeting the AQP1 gene and equivalent control constructs was carried out by Genecem (Shanghai Genecem Co., Ltd.). The AAVs were administered to the rodents via intracerebroventricular delivery into the lateral ventricle. The injection was performed using a stereotaxic device for guidance and began by drilling an aperture approximately 1 mm wide located 1 mm to the right and 0.3 mm in front of the bregma. Thereafter, 10 µl microsyringe needle was advanced via the aperture until it penetrated 2.5 mm into the lateral ventricle. Then, 5 µl of the AAV vector was injected into the ventricle at rate of 1 µl/min, after which the needle was slowly maintained in position for 2 min before removal. CSD modeling was performed three weeks after virus introduction [39].

Behavioral Testing

All behavioral assessments and related data analyses were performed in a blinded manner, with investigators unaware of group allocation throughout the experiments.

Novel Object Recognition Test (NORT)

The NORT was performed as described previously [40, 41], with slight modifications entailing three consecutive stages. The rodents were first acclimatized to the sidewall inside an enclosure (20 × 30 × 30 cm) for three days, allowing 3 min of free exploration. Twenty-four hours later, an object acclimation session was performed in which two identical items were arranged symmetrically within the enclosure, and the mice were allowed to explore freely for 5 min. Instances of exploration were defined as instances in which the nasal region came within ≤ 2 cm of an item. After an additional 24 h, one of the items was replaced with a new item that was a different color and shape. The animals were then returned to the enclosure for a 5-minute evaluation period, and engagement was documented with dedicated behavioral quantification software. The preference index was subsequently determined by calculating the proportion of time spent scrutinizing the new item relative to the total time spent scrutinizing familiar plus novel items.

Y-Maze Test

The Y-maze test was performed in accordance with previously documented protocols [42]. The setup for the Y-maze featured three extensions placed at 120° angles. Each rodent was assessed in two phases separated by a 1 h break. The preliminary phase began when the mouse was placed in the central position, after which time the mice were allowed to freely explore the maze excluding the unfamiliar extension for 5 min. In the second phase, each extension was unlocked, and the animals were placed in the original extension and allowed to freely explore all three extensions. Captured metrics included time spent in the unfamiliar extension and the number of entries. An entry was defined as when the rodent’s posterior limb crossed into an extension. This phase lasted 5 min.

Morris Water Maze (MWM) Test

The mice were subjected to five consecutive days of training before being subjected to an exploratory evaluation on the following day. A rounded basin 0.6 m in height and 1.2 m in width served as the experimental setup and was equipped with a concealed stand located in the southwestern sector. The duration needed by each animal to find the stand within the allotted 60-second test interval was recorded. On the fifth day, the stand was removed, and the animals were allowed to swim in the maze for 60 s. The captured data included the number of times the animals crossed the stand as well as the time spent within the designated sector [42].

RNA-Seq Data Processing and Analysis

RNA Isolation and Sequencing

Total RNA was extracted from hippocampal tissue samples of mice using TRIzol Reagent (Takara, Japan) following the manufacturer’s protocol. The concentration and quality of the RNA were evaluated using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). Only RNA samples that met the following quality control criteria were used: a 260/280 ratio between 1.8 and 2.2, OD260/230 ≥ 2.0, RNA Quality Number ≥ 6.5, a 28 S/18S ratio of ≥ 1.0, and a total RNA quantity of ≥ 1 µg. High-quality RNA was then used to construct the sequencing libraries according to standard protocols. High-throughput sequencing was performed to obtain comprehensive transcriptomic data.

Data Processing, Alignment, and Downstream Analyses

Raw reads (paired-end, 150 bp) were quality-filtered using fastp (v0.18.0) to remove adapter sequences and low-quality reads, generating clean reads. Residual rRNA reads were identified and removed by mapping to an rRNA database using Bowtie2 (v2.2.8). The remaining clean reads were aligned to the mouse reference genome using HISAT2 (default parameters), with gene annotations based on Ensembl release 110. Transcript assembly was performed using StringTie (v1.3.1). Gene-level raw counts were generated from aligned reads using featureCounts, and differential expression analysis was performed using DESeq2. For sample-level expression comparison and visualization, expression values were further normalized as transcripts per million (TPM). Genes with a |log2 fold change| > 1 and false discovery rate (FDR) < 0.05 were considered differentially expressed.

Principal component analysis (PCA) was performed using normalized expression data to evaluate overall sample clustering and assess potential batch effects. No obvious batch-related clustering was observed in the PCA plots, and samples were primarily grouped according to experimental conditions; therefore, no additional batch effect correction was applied.

Functional enrichment analyses of DEGs were conducted using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations. GO enrichment analysis was performed using the Goatools package, and KEGG pathway enrichment analysis was conducted using a Python-based pipeline (SciPy package for statistical testing). P values were corrected for multiple testing, and an adjusted P value (FDR) < 0.05 was considered statistically significant.

A per-sample summary of sequencing output, QC metrics (Q30, GC content, N bases), and alignment statistics (overall and uniquely mapped rates) is provided in Supplementary Table S3. This per-sample summary allows readers to evaluate sequencing quality, data processing, and alignment statistics for each sample.

Nissl Staining and HE Staining

Animal tissues were fixed in 4% paraformaldehyde and rinsed with 0.1 M PBS. A cryostat microtome was used to cut 10–20 μm sections. For Nissl staining, the sections were incubated in 1% toluidine blue solution for 15–30 min at 37 °C; thereafter, the sections were mounted with neutral balsam, differentiated, and cleared with xylene before examination by light microscopy. For HE staining, xylene was used for deparaffinization of the paraffin sections, which were then rehydrated through a graded series of alcohols. Hematoxylin staining was performed for 5–10 min, followed by differentiation in 1% hydrochloric acid‒alcohol solution, eosin staining for 3–5 min, and microscopic evaluation of cellular and tissue morphology to detect pathological alterations.

Immunofluorescence Staining

All immunofluorescence experiments were performed on frozen brain sections or cultured BV2 cells. For brain tissue preparation, tissues were collected following PBS perfusion, paraformaldehyde fixation, and sucrose dehydration, then cut into 10 μm thick frozen sections using a cryostat. Sections were permeabilized with Triton X-100, blocked with goat serum, and incubated overnight at 4 °C with primary antibodies. Subsequently, sections were incubated with appropriate Alexa Fluor–conjugated secondary antibodies, including Alexa Fluor 488–conjugated goat anti-rabbit IgG (H + L) (Abcam, UK; ab150077) and Alexa Fluor 594–conjugated goat anti-mouse IgG (H + L) (Abcam, UK; ab150116). DAPI (ab104139; Abcam, UK) was used for counterstaining the nuclei. Images were captured using a Nikon fluorescence microscope (Japan).

For BV2 cells, cells were grown on poly-L-lysine–coated coverslips, treated with LPS for 24 h, fixed with paraformaldehyde, rinsed with PBS, permeabilized with Triton X-100, and blocked with goat serum. Cells were incubated with primary antibodies at 4 °C overnight, followed by incubation with the same Alexa Fluor-conjugated secondary antibodies mentioned above, and DAPI staining for nuclear labeling. Images were acquired using a Nikon fluorescence microscope (Japan).

Immunohistochemistry

Immunohistochemistry was performed on frozen brain sections using a fluorescence-based detection method. Mice were anesthetized with isoflurane and perfused intracardially with PBS, followed by 4% paraformaldehyde. Brains were removed, post-fixed overnight at 4 °C, cryoprotected in sucrose solutions, embedded in OCT compound (Sakura Finetek), and sectioned into 20 μm thick frozen sections. Sections were permeabilized, blocked with PBS containing 5% donkey serum, and incubated with primary antibodies followed by Alexa Fluor–conjugated secondary antibodies, including Alexa Fluor 488–conjugated goat anti-rabbit IgG (H + L) (Abcam, UK; ab150077) and Alexa Fluor 594–conjugated goat anti-mouse IgG (H + L) (Abcam, UK; ab150116). Fluorescence images were captured using an Olympus VS120 Virtual Slide Scanner.

Cell Culture and Treatment

BV2 murine microglia were obtained from Procell in Wuhan, China. BV2 cells were cultured in low-glucose Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 1% penicillin/streptomycin and 10% fetal bovine serum (FBS; HyClone, Logan, UT, USA) at 37 °C in a humidified incubator containing 5% CO2. Cells at passages 3–6 were used for all experiments. AQP1 expression was silenced using three specific siRNAs (si-AQP1-1, si-AQP1-2, and si-AQP1-3), with a non-targeting siRNA serving as the negative control (siNC). When cells reached approximately 80% confluence, transfection was performed using Lipofectamine™ 3000 (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s protocol. After 6 h of incubation, the medium was replaced with fresh complete medium, and cells were cultured for an additional 48 h. Transfection efficiency was confirmed by RT‒qPCR and Western blotting. To establish the inflammatory model, BV2 cells were stimulated with LPS (1 µg/mL) for 24 h, a commonly used condition that reliably induces stable microglial activation and pro-inflammatory responses in BV2 cells, as previously reported [43–45]. The 24-h endpoint was used consistently for all BV2-based assays unless otherwise specified. For NLRP3 inflammasome activation, LPS-primed BV2 cells were subsequently treated with nigericin (10 µM) for 2 h, a widely used protocol reported to robustly activate NLRP3 inflammasome signaling in microglial models [46–48]. To further support the in vitro stimulation conditions used in this study, CCK-8-based viability assessment and condition-validation experiments were performed (Figure S4). To evaluate the effects of curcumin, BV2 cells were co-treated with LPS (1 µg/mL) and curcumin (10 µM) for 24 h. The optimal concentrations of both curcumin and Nig were determined in previous investigations [49].

Cell Viability Analysis

Cell viability was measured using the Cell Counting Kit-8 assay ((PF00004, Proteintech, China). Seed the cell suspension (100 µL/well) in a 96-well plate. Pre-incubate in a humidified incubator for a certain amount of time (37 °C, 5%CO2). Add 10 µL of CCK-8 solution to each well of the plate according to the instructions provided by the manufacturer and place the plate in the incubator for 1–4 h. Measure the absorbance of each well using an ELX808 microplate reader (Apbay Biotechnology (Suzhou) Co., Ltd.) at 450 nm. Viability levels of BV2 cells are normalized to a ratio of control groups.

Real-Time Quantitative Polymerase Chain Reaction (RT‒qPCR)

Total RNA was extracted from BV2 cells and mouse brain tissue with TRIzol reagent (Invitrogen, USA). The gene-specific primers used in this study were designed by Sangon Biotech (China) and are listed in Table 1. A One-Step RT reagent kit (Takara, China) was used to reverse transcribe 1 µg of total RNA. SYBR^® Premix Ex TaqTM II was used to conduct quantitative PCR in a 15 µL reaction on a LightCycler 96 system (Roche, Switzerland). β-Actin was used as an endogenous control, and each reaction was performed with technical triplicates. The relative mRNA expression levels were determined by the 2 − ΔΔCT method (Chang et al., 2025). The sequences of primers used for RT‒qPCR are listed in Supplementary Table 1.

Table 1.

Primer sequences (5’ to 3’)

Gene	Forward(5’−3’)	Reverse(5’−3’)
AQP1	CTGGCCTTTGGTTTGAGCAT	CCACACACTGGGCGATGAT
CD86	AGCACTATTTGGGCACAGAGAAAC	GTGAAGTCGTAGAGTCCAGTTGTTC
CD206	GCCACTGCCATGCCTACCAC	AGCTTGCCGTGCGTCTTGC
Arg1	CGGGGACCTGGCCTTTGTTG	TGGACCTCTGCCACCACACC
IL-10	TTAAGGGTTACTTGGGTTGC	GAGGGTCTTCAGCTTCTCAC
lL-4	CAGCTAGTTGTCATCCTGCTCTTC	GCCGATGATCTCTCTCAAGTGA
TNF-α	GGCATGGATCTCAAAGACAACC	CAGGTATATGGGCTCATACCAG
IL-1β	GAAATGCCACCTTTTGACAGTG	TGGATGCTCTCATCAGGACAG
iNOS	ACTCAGCCAAGCCCTCACCTAC	TCCAATCTCTGCCTATCCGTCTCG
β-actin	GATATCGCTGCGCTGGTCG	CATTCCCACCATCACACCCT

Open in a new tab

Enzyme-Linked Immunosorbent Assay (ELISA)

IL-18, IL-4, Arg1, IL-10, iNOS, TNF-α, NLRP3, ASC, pro-Caspase1, cleaved-Caspase1 and IL-1β levels were quantified using ELISA kits (Beijing 4 A Biotech Co., Ltd) according to the manufacturer’s instructions.

Western Blotting

Proteins from mouse hippocampal tissues and BV2 cells were isolated through the use of RIPA lysis buffer. The supernatant was collected after centrifugation at 12,000×g for 15 min at 4 °C, after which protein levels were assessed with a bicinchoninic acid quantification kit (TIANGEN, Beijing, China). Equal amounts of protein were subjected to electrophoretic separation via 10% SDS‒PAGE before transfer to polyvinylidene difluoride membranes (Millipore, Shenzhen, China). The membranes were blocked with 5% skim milk for 2 h, after which they were subjected to incubation with the primary antibody overnight at 4 °C. Thereafter, the membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (1:5000; Proteintech) at room temperature for 2 h. The protein signals were visualized via enhanced chemiluminescence. GAPDH functioned as the internal reference.

Virtual Screening

Virtual screening was performed with the L6000 natural product library (L6000-Targetmol Natural Product Library for HTS, encompassing 6509 compounds) obtained from TopScience. The 2D structure data files (SDF) were imported into Schrodinger software, after which 3D structures for individual compounds were generated with the LigPrep module employing the OPLS_4 force field. The Epik module was applied to calculate the protonation states of every feasible stereoisomer. The center of the docking grid was placed at the site of the crystallized ligand within the small-molecule GDP-binding pocket. Accommodation of the crystallized ligand necessitated modification of the outer box size, whereas the inner box was 10 Å. Molecular docking of the 3D structures prepared by LigPrep was performed with Schrödinger’s virtual screening workflow module. The virtual screening procedure involved four distinct phases. (1) HTVS: Generation of one conformer accompanied by retention of all stereoisomeric forms per ligand. The compounds with the highest scores (50%) were subjected to additional screening [30]. (2) SP: A single conformer was generated for each stereoisomer, and the top-scoring stereoisomer for every ligand was retained. The compounds in the top 20% were subjected to additional screening. (3) XP: One conformer was generated for only the highest-scoring stereoisomer per ligand, and the top 100 compounds were subjected to subsequent evaluation. (4) MM-GBSA: The binding free energies (ΔG bind) of the protein-ligand complexes were calculated using the Prime MM-GBSA module within the Schrödinger Suite 2025-1 (Maestro v14.3). The OPLS4 force field was employed for all calculations. The solvation effects were modeled using the VSGB 2.0 implicit solvation model, with the solvent dielectric constant set to 80.0 and the solute dielectric constant set to 1.0. Based on the best poses obtained from Glide XP docking, the protein-ligand complexes underwent local optimization (minimization). During this process, the ligand and protein residues within 5.0 Å of the ligand were allowed to relax (flexible), while the remaining protein atoms were constrained. The binding free energies were calculated using the following equation: ΔG bind=GComplex−(GProtein+GLigand), where G represents the total free energy, comprising molecular mechanics energy (EMM), polar solvation energy (GGB), and non-polar solvation energy (GSA). Conformational entropy contributions were not included in this calculation. Duplicate structures and compounds with less favorable predicted binding free energies (ΔG bind > − 50 kcal/mol) were excluded from further analysis.

Statistical Analyses

Data are presented as mean ± standard error of the mean (SEM) unless otherwise specified. The sample size (n) for each experiment, including behavioral, molecular, and in vitro assays, is indicated in the figure legends. All experiments were performed with appropriate biological replication, as indicated in the figure legends. Animals or samples were randomly assigned to experimental groups, and data analysis was conducted blinded to group identity whenever feasible.

Statistical analyses were performed using GraphPad Prism 9.5 (GraphPad Software, San Diego, CA, USA) and ImageJ for image quantification. For comparisons between two groups, an unpaired two-tailed Student’s t-test was applied. For comparisons involving more than two groups, one-way analysis of variance (ANOVA) was used, followed by Tukey’s post hoc test for multiple comparisons. P values < 0.05 were considered statistically significant. Exact P values are reported in figure legends, and significance is indicated as: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All statistical procedures, including randomization and blinding where applicable, were used to minimize bias and enhance reproducibility.

Results

Microglial M1 Polarization and Cognitive Changes Under CSD

To assess the impact of CSD on cognitive function, we first performed a series of behavioral tests. The experimental procedures are detailed in Fig. 1A. The Y-maze experiments revealed that compared with control mice, CSD mice exhibited significantly fewer entries and shorter dwell times in the novel arm (Fig. 1B, C), indicating impairments in spatial working memory. Novel object recognition (NORT) tests confirmed this finding, with CSD mice exhibiting a significant decrease in the novel object preference index (Fig. 1D), suggesting impairments in recognition memory. In the Morris water maze test, compared with control mice, CSD mice showed a significantly increased latency to reach the hidden platform and a reduced number of platform crossings in the probe trial (Fig. 1E, F), supporting CSD-induced cognitive impairment. Typical trajectories are shown in Fig. 1G-I. Given the critical role of the hippocampus in learning and memory, we further evaluated pathological damage to the hippocampus. Hematoxylin and eosin (HE) staining revealed that CSD mice had sparse neuronal arrangement and a reduced number of neurons in the hippocampus, along with nuclear condensation (Fig. 1J). Nissl staining confirmed a significant reduction in the number of Nissl bodies within the neuronal cell bodies, which was accompanied by disordered cell arrangement and cell body atrophy (Fig. 1J). These morphological changes provide structural evidence for the cognitive dysfunction induced by CSD.

Fig. 1 — CSD- associated microglial M1 polarization and cognitive impairment. A: Schematic of the animal experiment. B-F: Behavioral outcomes. B: Time (%) spent in the novel arm in the Y-maze. C: Number of entries into the novel arm in the Y-maze. D: NORT preference index. E: Latency to reach the hidden platform during acquisition trials. F: Number of platform crossings in the target quadrant during the MWM probe test. G-I: Representative trajectories. G: Y-maze movement tracks. H: NORT locomotor paths and occupancy; squares denote the novel object (NO) and circles denote the familiar object (FO); object identity and position were counterbalanced across animals. I: MWM spatial exploration trajectory during the probe test. J: Neuronal damage in the hippocampal region was assessed by HE staining and Nissl staining. Scale bar, 20 μm. K, L: Immunohistochemical analysis and quantification of IBA1 expression in hippocampal sections. Scale bar: 20 μm. M: Representative immunoblots showing IBA1, CD86, CD206, IL-1β, IL-6, and TNF-α protein levels in hippocampal tissue. GAPDH was used as a loading control. N: Quantification of the CD86/CD206 protein ratio based on densitometric analysis of the immunoblots. Band intensities were normalized to GAPDH, and data are presented as fold change relative to the control group. O: Densitometric quantification of immunoblot bands for inflammatory cytokines (IL-1β, IL-6, TNF-α) and the microglial marker IBA1. Band intensities were normalized to GAPDH and are shown as fold change relative to the control group. Statistical analysis: For quantitative comparisons between Control and CSD groups (B-F, L, N, O), data were analyzed using an unpaired two-tailed Student’s t-test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For behavioral tests (B-F), each dot represents one mouse (n = 10 mice/group). For immunohistochemical quantification (L) and immunoblot densitometry (N and O), each dot represents one mouse (n = 3 mice/group)

To elucidate the potential mechanisms underlying hippocampal neuronal damage, we investigated neuroinflammatory responses and microglial activation. Neuroinflammation, particularly microglial activation, plays a crucial role in the onset and progression of neurodegenerative diseases and cognitive dysfunction [50, 51]. Microglia, the resident immune cells of the central nervous system, typically become activated and polarize into different functional phenotypes in response to pathological stimuli, thereby initiating neuroinflammatory responses [52]. Immunohistochemical analysis revealed a significant increase in the number of IBA1-positive microglia in the hippocampus of CSD-treated mice (Fig. 1K, L), indicating widespread microglial activation. Given the central role of microglia in immune surveillance within the central nervous system and their critical involvement in initiating neuroinflammatory responses [53], we analyzed the impact of CSD on microglial function. The extensive microglial activation induced by CSD led to their polarization toward the proinflammatory M1 phenotype, a key step in neuroinflammation. Western blot analysis revealed significant upregulation of the M1 microglial marker CD86 and downregulation of the expression of the M2 marker CD206 in the CSD group, indicating that CSD promotes microglia polarization toward the M1 phenotype. M1 microglia release many proinflammatory cytokines, thereby activating neuroinflammatory responses. Consistent with these findings, the levels of proinflammatory cytokines (such as IL-1β, TNF-α, and IL-6) in the hippocampus were significantly elevated in the CSD group (Fig. 1M-O). These results suggest that CSD activates microglia, promoting their polarization toward the M1 phenotype, which in turn triggers neuroinflammatory responses that may contribute to hippocampal neuronal damage and cognitive decline.

AQP1 Expression and Microglial Localization Under CSD

To elucidate the molecular mechanisms underlying the cognitive impairment induced by CSD, we performed a transcriptomic sequencing analysis on mouse hippocampal tissues. Principal component analysis (PCA) of the hippocampal RNA-seq profiles revealed tight within-group clustering and clear separation between the control and CSD samples, primarily along PC1 (explained 97.61% of the variance; PC2, 2.11%) (Fig. 2A). On the basis of the effect size and significance thresholds of |log2FC|>1 and adjusted P value FDR < 0.05, the volcano plot revealed 912 upregulated genes and 570 downregulated genes (Fig. 2B). A gene expression heatmap revealed significant upregulation of genes such as AQP1, Folr1, Scara5, Steap1, Clic6, and Slc28a3, whereas genes such as Ddit4, Dio2, Arc, Akr1c18, and Glt8d2 were significantly downregulated (Fig. 2C). KEGG pathway analysis further revealed the NOD-like receptor signaling pathway as one of the most significantly altered pathways (Fig. 2D), indicating its potential central role in CSD-induced neuroimmune modulation. Gene Ontology (GO) enrichment analysis revealed that the upregulated genes were predominantly associated with immune cell activation and cytokine release (Fig. 2E), suggesting that immune responses play a key role in the neuroinflammation triggered by CSD.

Fig. 2 — Expression of AQP1 after CSD. A: Principal component analysis (PCA) of the hippocampal RNA-seq profiles revealed tight within-group clustering and clear separation between the Control group and the CSD group along PC1 (variance explained: 97.61% for PC1 and 2.11% for PC2; n = 4/group). B: Volcano plot of differentially expressed genes (DEGs) between the CSD and Control groups. Dashed lines denote preset thresholds (|log2FC|>1; FDR < 0.05); upregulated genes are shown in red, and downregulated genes are shown in blue. C: Heatmap of representative DEGs whose expression significantly changed across samples. D: KEGG enrichment dot plot of the DEGs. E: GO biological process enrichment dot plot of the DEGs. F: The expression levels of upregulated genes in hippocampal samples from control and CSD mice were measured by RT‒qPCR. G: Representative immunoblots showing AQP1 protein levels at the indicated time points following chronic sleep deprivation (CSD) treatment. GAPDH was used as a loading control. H: Densitometric quantification of AQP1 protein expression from panel G. Band intensities were normalized to GAPDH and are presented as fold change relative to the control group. I: RT‒qPCR measurement of relative AQP1 mRNA expression at the indicated time points. Statistical analysis: RNA-seq differential expression was performed using DESeq2, and DEGs were defined as |log2FC| > 1 and FDR < 0.05 (A-E; n = 4 mice/group). For RT-qPCR validation between Control and CSD groups (F), data were analyzed using an unpaired two-tailed Student’s t-test (each dot represents one mouse; n = 3 mice/group). For the time-course analyses of AQP1 protein (H) and AQP1 mRNA expression (I) across multiple time points, one-way ANOVA followed by Tukey’s post hoc test was applied (each dot represents one mouse; n = 3 mice/time point). Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; n.s., not significant

Among all the upregulated genes, AQP1 exhibited the most significant increase in expression, and previous studies have shown that AQP1 plays a critical role in microglial polarization (Fig. 2F). Therefore, we selected AQP1 as the core candidate gene for further validation. qPCR and Western blotting revealed that AQP1 expression increased at both the mRNA and protein levels in a time-dependent manner after CSD treatment, peaking at 24 h (Fig. 2G–I). Immunofluorescence colocalization analysis confirmed that AQP1 was predominantly expressed in microglia, with high colocalization with IBA1-positive microglia but minimal colocalization with GFAP-expressing astrocytes or NeuN-expressing neurons (Fig. 3A–F). These results suggest that the upregulation of AQP1 expression may modulate microglia activation, contributing to CSD-induced neuroinflammation in the hippocampus and cognitive impairment.

Fig. 3 — Localization of AQP1 after CSD. A, B: Representative immunofluorescence staining images of IBA1 (green) and AQP1 (red) expression in hippocampal slices from control and CSD mice. Scale bars: 200 and 20 μm. C, D: Representative immunofluorescence staining images of GFAP (green) and AQP1 (red) expression in hippocampal slices from control and CSD mice. Scale bars: 200 and 20 μm. E, F: Representative immunofluorescence staining images of NeuN (green) and AQP1 (red) expression in hippocampal slices from control and CSD mice. Scale bars: 200 and 20 μm. Statistical analysis: Quantitative comparisons between Control and CSD groups (B, D, F) were analyzed using an unpaired two-tailed Student’s t-test. Data are presented as mean ± SEM. For immunofluorescence quantification, each dot represents one mouse (n = 3 mice/group). n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Effects of AQP1 Knockdown On Cognitive Outcomes and Hippocampal Structure Under CSD

To evaluate the role of AQP1 in the cognitive dysfunction induced by CSD (Fig. 4A), we established AQP1 knockdown expression in the hippocampus through intracerebroventricular injection of AAV-shRNA in the mouse model (Fig. 4B-F). We then assessed the effects using a series of behavioral tests. The Y-maze test results revealed that CSD mice exhibited significantly fewer entries and less time spent in the novel arm, indicating impairments in spatial working memory. AQP1 knockdown significantly improved spatial exploration and working memory deficits (Fig. 4G, H). In the NORT, CSD mice showed a significant decrease in the preference index for the novel object, and AQP1 knockdown partially restored their recognition ability (Fig. 4I). Furthermore, in the Morris water maze test, CSD mice exhibited a significantly longer latency to reach the hidden platform and fewer platform crossings in the probe trial; these deficits were significantly alleviated by AQP1 knockdown. In contrast, the time spent in the target quadrant did not differ significantly among groups (Fig. 4J–L). The behavioral trajectories are shown in Fig. 4M-O. These results indicate that AQP1 knockdown significantly ameliorates the spatial learning and recognition memory deficits induced by CSD.

To investigate the cellular and histological basis of the improvement in cognition following AQP1 knockdown, we assessed structural damage to the hippocampus induced by CSD. Histopathological analysis revealed significant neuronal damage, including nuclear condensation, cell body shrinkage, and a reduction in the number of Nissl bodies, in the hippocampus of CSD mice. However, AQP1 knockdown significantly alleviated these pathological changes (Fig. 4P). TUNEL staining further demonstrated that CSD significantly increased the proportion of apoptotic cells in the hippocampus, whereas AQP1 knockdown significantly reduced apoptosis (Fig. 4Q, R). These results suggest that AQP1 plays a critical role in hippocampal pathology induced by CSD and that its knockdown significantly alleviates both structural and functional damage to the hippocampus, providing further support for AQP1 as a potential therapeutic target for CSD-related cognitive dysfunction.

AQP1 is Associated with Microglial Polarization and Inflammatory Responses Under CSD

Cognitive impairment induced by chronic sleep deprivation (CSD) is closely associated with neuroinflammation, in which microglial activation plays a central role [54]. Under pathological conditions, microglia polarize toward proinflammatory M1 or anti-inflammatory M2 phenotypes, thereby shaping the neuroinflammatory milieu [55]. To determine whether AQP1 is associated with microglial polarization under CSD, immunofluorescence staining was performed to assess IBA1-positive microglia and their phenotype markers in hippocampal tissue. Compared with control mice, CSD markedly increased IBA1 expression and the M1 marker CD86, while significantly reducing the M2 marker CD206. Notably, AQP1 knockdown attenuated these changes, as evidenced by decreased CD86 and increased CD206 expression (Fig. 5A–E). Consistently, Western blot analysis showed that CSD significantly elevated the CD86/CD206 protein ratio, whereas AQP1 knockdown markedly reduced this ratio, consistent with a shift toward the M2 phenotype (Fig. 5F, G).

Fig. 5 — AQP1 knockdown is associated with an M2-like shift in microglial polarization in the hippocampus of CSD mice. A-E: Immunofluorescence staining images of hippocampal sections labeled for IBA1 (microglia), CD86 (M1 marker), and CD206 (M2 marker), with DAPI indicating the nuclei, followed by quantification of the fluorescence signals. Scale bar: 200 μm. F: Representative immunoblots showing CD86 and CD206 protein levels in different experimental groups. GAPDH was used as a loading control. G: Densitometric quantification of the CD86/CD206 protein ratio from panel F. Band intensities were normalized to GAPDH and are presented as fold change relative to the control group. Statistical analysis: For multiple-group comparisons (C, D, and G), data were analyzed using one-way ANOVA followed by Tukey’s multiple-comparisons test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For immunofluorescence quantification (C and D) and immunoblot densitometry (G), each dot represents one mouse (n = 3 mice/group)

Supporting these observations, qPCR and ELISA analyses demonstrated that AQP1 knockdown increased M2-associated markers while suppressing M1-related proinflammatory factors (Figure S1A–L). Collectively, these results support an association between AQP1 and microglial polarization under CSD conditions.

To further explore the mechanisms potentially involved in the effects of AQP1 on microglial polarization, we next focused on the NLRP3 inflammasome pathway, which was highlighted as a candidate pathway from transcriptomic analysis. Western blotting revealed that CSD was associated with increased NLRP3 inflammasome–related readouts in the hippocampus, as indicated by increased expression of NLRP3, ASC, and cleaved caspase-1. Importantly, these changes were significantly attenuated by AQP1 knockdown (Fig. 6A–E). Immunohistochemical staining further confirmed that AQP1 deficiency was associated with attenuated CSD-induced upregulation of NLRP3 and ASC (Fig. 6F–H). In parallel, ELISA analysis showed that CSD significantly increased IL-1β and IL-18 levels, whereas AQP1 knockdown markedly reduced their expression (Figure S1M, N). Together, these findings suggest that AQP1 may contribute to CSD-induced neuroinflammation, at least in part, through effects on NLRP3 inflammasome–related inflammatory responses.

Fig. 6 — AQP1 knockdown is associated with reduced NLRP3 inflammasome–related inflammatory readouts in the hippocampus of CSD mice. A: Representative immunoblots showing hippocampal protein levels of NLRP3, ASC, pro-caspase-1, cleaved caspase-1, and IL-1β in the indicated groups (Control, CSD, AAV-sh-NC + CSD, and AAV-sh-AQP1 + CSD). GAPDH was used as a loading control. B–E: Densitometric quantification of NLRP3 (B), ASC (C), the cleaved caspase-1/pro-caspase-1 ratio (D), and IL-1β (E) from panel A. Band intensities were normalized to GAPDH where applicable, and data are presented as fold change relative to the control group. F–H: An immunohistochemical staining assay was performed to measure the NLRP3 and ASC levels in the hippocampal tissues. Scale bars: 200 and 20 μm. Statistical analysis: For multiple-group comparisons (B-E, F, and G), data were analyzed using one-way ANOVA followed by Tukey’s multiple-comparisons test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For immunoblot densitometry (B-E) and immunohistochemical quantification (F and G), each dot represents one mouse (n = 3 mice/group)

NLRP3 Signaling is Associated with AQP1-Related Microglial Polarization and Cognitive Changes Under CSD

To further assess the contribution of NLRP3 signaling to AQP1-associated effects, the NLRP3 agonist nigericin (Nig) was administered. Behavioral assessments, including the Y-maze, novel object recognition, and Morris water maze tests, showed that AQP1 knockdown significantly improved cognitive performance in CSD mice. In contrast, Nig treatment markedly attenuated these improvements in AAV-sh-AQP1 + CSD mice, suggesting that nigericin (an NLRP3 agonist) attenuates the cognitive improvements observed with AQP1 knockdown (Fig. 7A–H).

Fig. 7 — NLRP3 signaling is associated with AQP1-related microglial polarization and cognitive changes under CSD. Groups: Control, CSD, AAV-sh-NC + CSD, AAV-sh-AQP1 + CSD, and AAV-sh-AQP1 + Nig (nigericin) + CSD. A-E: Behavioral tests. A: Time (%) spent in the novel arm in the Y-maze. B: Number of entries into the novel arm in the Y-maze. C: NORT discrimination index. D: Latency (s) to reach the hidden platform during the MWM acquisition trials. E: Number of platform crossings in the target quadrant during the MWM probe test. F-H: Representative trajectories. F: Y-maze movement tracks. G: NORT locomotor paths/occupancy; squares denote the novel object (NO), and circles denote the familiar object (FO); object identity and position were counterbalanced across animals. H: MWM probe trajectories relative to the former platform location. I, J: Representative immunohistochemical staining images of CD86 and CD206 in hippocampal sections; scale bar: 200 μm. K: Representative immunoblots showing inflammasome-related proteins (NLRP3, ASC, pro-caspase-1, cleaved caspase-1, and IL-1β) in hippocampal tissue. GAPDH was used as a loading control. L: Densitometric quantification of NLRP3, ASC, cleaved caspase-1, and IL-1β bands from panel K. Band intensities were normalized to GAPDH where applicable, and data are presented as fold change relative to the control group. Statistical analysis: For multiple-group comparisons (A–E, J, and L), data were analyzed using one-way ANOVA followed by Tukey’s multiple-comparisons test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For behavioral tests (A–E), each dot represents one mouse (n = 10 mice/group). For immunohistochemical quantification (J) and immunoblot densitometry (L), each dot represents one mouse (n = 3 mice/group)

Next, we explored whether nigericin treatment is associated with changes in the microglial polarization changes associated with AQP1 knockdown. Immunohistochemical analysis of hippocampal sections from CSD mice indicated that hippocampal microglia shifted showed an M1-like shift with reduced M2-like features (CD86↑ and CD206↓). AQP1 knockdown was associated with a shift toward an M2 phenotype, whereas this effect was significantly attenuated by nigericin treatment (Fig. 7I, J). Consistent with these observations, qPCR and ELISA analyses showed that AQP1 knockdown increased the expression of M2-associated markers and reduced M1-related proinflammatory factors, whereas NLRP3 activation largely attenuated these effects (Figure S2A–I).

At the molecular level, Western blot analysis further showed that Nig treatment was associated with increased NLRP3 inflammasome–related protein expression under AQP1 knockdown conditions, as reflected by increased levels of NLRP3, ASC, cleaved caspase-1, and IL-1β (Fig. 7K, L). Collectively, these findings suggest that the protective effects associated with AQP1 knockdown in CSD mice are accompanied by changes in NLRP3 inflammasome–related inflammatory responses and microglial polarization, and that NLRP3 signaling is likely functionally relevant to these effects.

AQP1 is Associated with Microglial Polarization and NLRP3 Inflammasome–Related Inflammatory Responses in BV2 Cells

To further investigate the role of AQP1 in microglia, we established a stable BV2 microglia line with low AQP1 expression, and the knockdown efficiency was confirmed by RT‒qPCR, Western blotting, and immunofluorescence (Figure S3A‒E). si-AQP1-1 was subsequently selected for further analysis. To model neuroinflammatory responses, the cells were treated with lipopolysaccharide (LPS), after which the impact of AQP1 knockdown on microglial polarization was evaluated. Previous in vivo experiments demonstrated that CSD induces microglial polarization toward the proinflammatory M1 phenotype and triggers neuroinflammation. On this basis, we investigated the impact of LPS treatment on microglial polarization in vitro. Western blotting revealed that LPS treatment significantly increased the CD86/CD206 protein ratio, consistent with an M1-like shift. AQP1 knockdown attenuated this change and was associated with an M2-like shift (increased CD206-positive cells) (Fig. 8A, B). These results suggest that AQP1 knockdown is associated with reduced M1-like polarization and enhanced M2-like features. ELISA confirmed these findings, showing that LPS treatment significantly increased the expression of M1 markers (IL-1β, iNOS, and TNF-α) and decreased the expression of M2 markers (Arg1, IL-10, and IL-4), whereas AQP1 knockdown attenuated these changes by decreasing the expression of proinflammatory factors and upregulating the expression of anti-inflammatory factors (Fig. 8C–H).

Fig. 8 — AQP1 knockdown is associated with changes in microglial polarization and NLRP3 inflammasome–related inflammatory responses in BV2 cells. A: Representative immunoblots showing CD86 and CD206 protein levels in BV2 cells under the indicated treatments. GAPDH was used as a loading control. B: Densitometric quantification of the CD86/CD206 protein ratio based on panel A. Band intensities were normalized to GAPDH, and data are presented as fold change relative to the control group. C–E: ELISA of M1-associated protein (TNF-α, IL-1β, iNOS) expression in BV2 cell culture supernatants. F–H: ELISA of M2-associated protein (IL-4, IL-10, and Arg1) expression in BV2 cell culture supernatants. I: Representative immunoblots showing inflammasome-related proteins (NLRP3, ASC, pro-caspase-1, cleaved caspase-1 p20, and IL-1β) in BV2 cells under the indicated treatments. GAPDH was used as a loading control. J: Densitometric quantification of NLRP3, ASC, cleaved caspase-1 (p20), and IL-1β bands from panel I. Band intensities were normalized to GAPDH where applicable and are presented as relative protein expression (fold change) relative to the control group. K‒M: The expression levels of NLRP3 and ASC in BV2 cells were detected by immunofluorescence staining. Scale bar: 20 μm. Statistical analysis: For multiple-group comparisons (B–H, J, L, and M), data were analyzed using one-way ANOVA followed by Tukey’s multiple-comparisons test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 3 independent biological replicates (independent cell culture experiments) for all quantitative analyses

We next examined NLRP3 inflammasome–related inflammatory responses in this model. Western blot analysis showed that LPS treatment increased the expression of NLRP3, ASC, cleaved caspase-1, and IL-1β, consistent with increased NLRP3 inflammasome–related readouts under LPS stimulation. AQP1 knockdown was associated with reduced expression of these proteins (Fig. 8I, J). Immunofluorescence staining of NLRP3 and ASC showed a similar trend (Fig. 8K–M).

To further assess the functional relevance of NLRP3 signaling may affect AQP1-associated microglial polarization changes, we performed an intervention experiment using the NLRP3 agonist nigericin (Nig). Immunofluorescence staining showed that Nig treatment partially attenuated the decrease in CD86-positive cells and the increase in CD206-positive cells observed after AQP1 knockdown (Fig. 9A–C). Western blot analysis further showed that Nig treatment markedly attenuated the reduction in NLRP3, ASC, cleaved caspase-1, and IL-1β expression associated with AQP1 knockdown, with protein levels shifting toward those observed in the LPS-treated group (Fig. 9D, E).

Fig. 9 — Nigericin attenuates AQP1 knockdown–associated changes in microglial polarization and NLRP3 inflammasome–related inflammatory responses in BV2 cells. A-C: Representative immunofluorescence staining images of the M1 marker CD86 and M2 marker CD206 in BV2 cells under five experimental conditions (control, LPS, si-NC + LPS, si-AQP1 + LPS, and si-AQP1 + Nig + LPS). Scale bar: 50 μm. D: Representative immunoblots showing inflammasome-related proteins (NLRP3, ASC, pro-caspase-1, cleaved caspase-1 p20, and IL-1β) in BV2 cells under the indicated treatments. GAPDH was used as a loading control. E: Densitometric quantification of NLRP3, ASC, cleaved caspase-1 (p20), and IL-1β bands from panel D. Band intensities were normalized to GAPDH where applicable and are presented as fold change relative to the control group. Statistical analysis: Quantitative data (C and E) were analyzed using one-way ANOVA followed by Tukey’s multiple-comparisons test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 3 independent biological replicates (independent cell culture experiments) for all quantitative analyses

In summary, in the BV2 cell model, AQP1 knockdown was associated with changes in microglial polarization and NLRP3 inflammasome–related inflammatory responses under LPS stimulation. Pharmacological activation of NLRP3 weakened these effects, suggesting that NLRP3 signaling may contribute to the AQP1-associated inflammatory phenotype in vitro.

Curcumin is Associated with AQP1 Expression Changes Under CSD

Based on the observed involvement of AQP1 in CSD-induced cognitive impairment, we further explored the potential of AQP1-associated intervention strategies by screening active compounds. Through high-throughput virtual screening of a natural small-molecule library (Fig. 10A), curcumin showed favorable predicted binding affinity to AQP1, with an predicted MM-GBSA binding free energy of −63.26 kcal/mol (Fig. 10B–D), suggesting a putative in silico interaction between curcumin and AQP1.

Fig. 10 — Curcumin is associated with AQP1 expression and neuroinflammatory changes under CSD. A: Virtual screening workflow for AQP1 ligands. B-D: Molecular docking results showing the binding of curcumin to AQP1. E: The number of hydrogen bonds between curcumin and AQP1 during the molecular dynamics simulations. F: RMSD of the curcumin–AQP1 complex during the simulations. G: The radius of gyration (Rg) of the curcumin–AQP1 complex during the simulation. H: Representative immunoblots showing AQP1 and IL-1β protein levels in hippocampal tissue from the indicated groups (Control, CSD, and Curcumin + CSD). GAPDH was used as a loading control. I: Representative immunoblots showing IL-6 and TNF-α protein levels in hippocampal tissue from the indicated groups. GAPDH was used as a loading control. J: Densitometric quantification of AQP1, IL-1β, IL-6, and TNF-α bands from panels H and I. Band intensities were normalized to GAPDH and are presented as fold change relative to the control group. Statistical analysis: Quantitative data in panel J were analyzed using one-way ANOVA followed by Tukey’s multiple-comparisons test. Data are presented as mean ± SEM. n.s., not significant; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For immunoblot densitometry (J), each dot represents one mouse (n = 3 mice/group)

To further assess the stability of the predicted curcumin–AQP1 interaction, we performed molecular dynamics simulations. The simulation results showed relatively limited RMSD fluctuation of the curcumin–AQP1 complex during the simulation period, along with persistent hydrogen-bonding interactions and stable solvent-accessible surface characteristics (Fig. 10E, F), supporting a relatively stable predicted binding conformation. Free energy landscape analysis further indicated that the curcumin–AQP1 complex occupied a low free-energy state (Fig. 10G), supporting a plausible predicted binding mode.

To further examine the biological effects of curcumin under CSD conditions, Western blot analysis was performed in hippocampal tissue. Curcumin treatment was associated with reduced expression of AQP1, IL-1β, IL-6, and TNF-α in CSD mice (Fig. 10H–J). These findings support the potential neuroprotective potential of curcumin under CSD conditions; however, direct target engagement/binding specificity between curcumin and AQP1 remains to be established and will require dedicated validation in future studies.

Curcumin is Associated with Reduced Neuroinflammatory Responses and Improved Behavioral Outcomes

To evaluate the effects of curcumin on CSD-induced cognitive dysfunction (Fig. 11A), we first performed a series of behavioral tests. CSD mice treated with curcumin showed significant improvements in the Y-maze, novel object recognition, and Morris water maze tests. Specifically, compared with untreated CSD mice, curcumin-treated mice spent more time exploring the novel arm, showed higher preference indices for the novel object, and exhibited improved spatial memory performance (Fig. 11B–I). These results suggest that curcumin treatment improves behavioral performance in CSD mice.

At the molecular level, Western blot analysis showed that curcumin treatment was associated with reduced AQP1 expression in the hippocampal tissue of CSD mice and a decreased CD86/CD206 protein ratio (Fig. 11J, K). These findings suggest that curcumin treatment is associated with a shift in microglial polarization away from the proinflammatory M1-like phenotype and toward an M2-like phenotype, accompanied by reduced neuroinflammatory markers. In addition, curcumin treatment was associated with reduced NLRP3 inflammasome–related readouts under CSD conditions, as reflected by reduced expression of NLRP3, ASC, and cleaved caspase-1 (Fig. 11L, M).

In the LPS-induced BV2 microglial inflammation model, curcumin treatment showed a similar pattern, with reduced M1-like polarization and reduced NLRP3 inflammasome–related readouts (Fig. 11N–R), consistent with curcumin-associated effects in the in vitro model, In summary, curcumin treatment improved behavioral performance and was associated with changes in AQP1 expression, microglial polarization, and NLRP3 inflammasome–related inflammatory responses in vivo and in vitro. These findings support the potential neuroprotective value of curcumin in CSD-associated neuroinflammation and cognitive dysfunction, while further studies are needed to clarify the direct target engagement/binding specificity between curcumin and AQP1.

Discussion

The present study provides integrated in vivo and in vitro evidence supporting the involvement of AQP1-associated inflammatory changes in chronic sleep deprivation (CSD)-induced neuroinflammation and cognitive dysfunction. By combining behavioral, histopathological, transcriptomic, and molecular analyses, our findings suggest that AQP1 as an important factor associated with microglia-mediated inflammatory responses under CSD conditions and is associated with CSD-related neuroinflammation. In addition, curcumin intervention alleviated CSD-associated cognitive impairment and neuroinflammatory responses, accompanied by reduced AQP1 expression and a shift toward a less proinflammatory microglial phenotype. Collectively, these findings support the involvement of AQP1-associated inflammatory pathways in CSD-related neuroinflammation and cognitive dysfunction, and support further evaluation of curcumin in CSD-related neuroinflammation, while direct target engagement and specificity remain to be established (Fig. 12).

Fig. 12 — Schematic model of the proposed pathway. Chronic sleep deprivation (CSD) is associated with increased AQP1 expression in the hippocampus, accompanied by microglial polarization toward a pro-inflammatory M1-like phenotype and enhanced NLRP3 inflammasome–related responses. These changes are associated with increased inflammasome-related readouts, including caspase-1 cleavage and elevated IL-1β/IL-18 levels, consistent with aggravated neuroinflammation. Under CSD conditions, curcumin treatment is associated with reduced AQP1 expression and reduced NLRP3 inflammasome–related readouts, accompanied by a shift toward a less pro-inflammatory microglial phenotype, accompanied by an overall reduction in neuroinflammatory changes. The direct molecular relationship between curcumin and AQP1 requires further validation

Sleep is essential for maintaining neural homeostasis and cognitive function, and chronic sleep loss is increasingly recognized as a major risk factor for neurological and systemic disorders [56–58]. Consistent with previous studies [59–61], our behavioral results (Y-maze, NORT, and Morris water maze) showed that CSD induced significant impairments in spatial working memory, recognition memory, and spatial learning ability. These behavioral abnormalities were accompanied by structural damage in the hippocampus, including sparse neuronal arrangement, nuclear pyknosis, and reduced Nissl bodies, particularly in the CA3 region. Together, these observations support the view that CSD is associated with both functional and structural deficits in the hippocampus.

Neuroinflammation, especially microglial activation and polarization, is widely regarded as a contributing mechanism underlying CSD-related cognitive impairment. Microglia are highly plastic immune cells that can adopt either a proinflammatory (M1-like) or anti-inflammatory (M2-like) phenotype in response to environmental cues. Excessive or sustained M1-like polarization amplifies inflammatory signaling and exacerbates neuronal injury, whereas M2-like polarization contributes to inflammation resolution and tissue repair [62, 63]. In the present study, CSD markedly increased microglial activation and was associated with an (M1-like) polarization shift, characterized by increased CD86 expression and elevated proinflammatory cytokine production. Importantly, AQP1 knockdown was associated with reduced M1-associated markers and increased M2-associated markers, suggesting that AQP1 may contribute to microglial polarization changes during CSD-associated neuroinflammation.

Transcriptomic analysis identified AQP1 as one of the most prominently upregulated genes in the hippocampus after CSD exposure, and pathway enrichment analyses highlighted immune activation and NOD-like receptor-related signaling. Subsequent qPCR, Western blotting, and immunofluorescence analyses consistently confirmed increased AQP1 expression in hippocampal tissue and its association with microglia-related inflammatory changes. Functional experiments further showed that AQP1 knockdown improved behavioral performance, reduced hippocampal neuronal injury, and was associated with a shift in microglial polarization toward a less pro-inflammatory (M1-like) phenotype. These findings support an association between AQP1 and neuroinflammatory responses under CSD conditions. Our data also indicate that AQP1 knockdown was accompanied by reduced NLRP3 inflammasome-related readouts, including decreased levels of NLRP3, ASC, cleaved caspase-1, and IL-1β, in parallel with reduced pro-inflammatory polarization. Moreover, pharmacological activation of NLRP3 with nigericin attenuated the anti-inflammatory and protective effects of AQP1 knockdown in both in vivo and in vitro settings. Together, these results support an association between AQP1-associated inflammatory changes and NLRP3 inflammasome–related signaling in our models.

Although the present study was not designed to define the direct direct molecular relationship between AQP1 and NLRP3 activation, several biologically plausible mechanisms may underlie the observed association. First, AQP1 may influence intracellular ionic homeostasis and membrane dynamics, which could affect K⁺ efflux, a well-established trigger of NLRP3 inflammasome activation [64–66]. Second, altered AQP1 expression may contribute to oxidative stress and mitochondrial dysfunction, both of which are closely linked to inflammasome activation in microglia [67–69]. Third, AQP1-related changes in cell volume regulation and membrane permeability may influence inflammatory signaling thresholds during chronic stress [70]. These possibilities are consistent with the overall pattern of our findings and prior literature; however, they remain speculative in the context of CSD and require direct mechanistic testing in future studies.

It is also important to distinguish the potential role of AQP1 from that of AQP4 in neuroinflammation. AQP4 has been more extensively studied in astrocytes, particularly in relation to water homeostasis, cerebral edema, and glymphatic function [71–73]. In addition, emerging evidence suggests that AQP1 may be more closely associated with inflammatory regulation in specific cellular contexts, including microglia-related responses [19, 74, 75]. Consistent with this, the present findings further support a potential role for AQP1 in microglia-associated inflammatory changes under CSD conditions. This comparison does not imply that AQP1 and AQP4 act in mutually exclusive pathways; rather, it highlights the likelihood that different aquaporin isoforms contribute to neuroinflammation through cell type- and context-dependent mechanisms. A more detailed dissection of isoform-specific and cell-specific functions will be important in future work.

Curcumin is a pleiotropic natural compound with reported antioxidant, anti-inflammatory, and neuroprotective effects [76–78]. In the present study, curcumin improved behavioral outcomes in CSD mice and attenuated pro-inflammatory microglial polarization and inflammasome-related responses in both hippocampal tissue and LPS-stimulated BV2 cells. These changes occurred alongside with reduced AQP1 expression and are consistent with modulation of AQP1-related inflammatory phenotypes. However, given the broad range of signaling pathways influenced by curcumin, including NF-κB and Nrf2 pathways that are tightly linked to neuroinflammation and oxidative stress [79–81], it is likely that curcumin exerts its effects through multiple mechanisms. Accordingly, in this study, curcumin is discussed as a pharmacological probe, and direct target engagement/binding specificity for AQP1 was not established.

Several limitations of the present study should be acknowledged. First, although our data support a functional association between AQP1 expression and NLRP3 inflammasome–related responses, direct molecular interactions and upstream regulatory mechanisms were not examined in this study; therefore, causal mechanistic links cannot be conclusively established. Future studies could explore direct interaction assays and upstream pathways to further elucidate how AQP1 relates to NLRP3 signaling. Second, EEG/EMG-based polysomnography was not performed; therefore, sleep architecture (e.g., NREM/REM staging and sleep-fragmentation indices) could not be directly quantified, The CSD paradigm was implemented according to previously validated protocols. Future studies incorporating EEG/EMG recordings would help provide direct sleep-stage validation and quantify sleep fragmentation under this CSD protocol. Third, BV2 cells were used as a mechanistically tractable in vitro model to complement our in vivo findings; however, as an immortalized microglial line, they may not fully recapitulate the heterogeneity and functional complexity of primary microglia in vivo. Future work incorporating primary microglia and microglia-targeted genetic approaches would help further validate and refine the mechanistic interpretation. Finally, our current study does not establish direct target engagement or binding specificity between curcumin and AQP1, and causal mechanisms remain untested. Future studies incorporating dedicated target-engagement and specificity assays (e.g., biochemical binding assays, CETSA/DARTS, or genetic rescue/epistasis experiments) would help to clarify whether curcumin directly binds to AQP1. Additionally, physiological parameters such as body weight, food intake, locomotor activity, and endocrine/metabolic parameters were not systematically quantified as formal study endpoints. Future studies should include a more comprehensive assessment of these parameters to refine the understanding of chronic sleep deprivation’s systemic effects on health and behavior.

Conclusion

In conclusion, this study indicates that CSD is associated with AQP1 upregulation, proinflammatory (M1-like) microglial polarization, and enhanced inflammasome-related inflammatory responses, which are accompanied by neuroinflammation and cognitive dysfunction. AQP1 knockdown attenuated these changes. Curcumin treatment improved behavioral and inflammatory outcomes, occurring alongside reduced AQP1 expression; whether AQP1 is a direct molecular target requires further validation. Together, these findings provide a working framework for understanding CSD-related neuroinflammation and support further investigation of AQP1-associated inflammatory pathways a potentially relevant therapeutic direction.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary figure 13^{(3.8MB, png)}

(PNG 3.75 MB)

Supplementary Material 1 (TIF 7.32 MB)^{(7.3MB, tif)}

Supplementary figure 14^{(5.9MB, png)}

(PNG 5.91 MB)

Supplementary Material 2 (TIF 3.10 MB)^{(3.1MB, tif)}

Supplementary figure 15^{(11.4MB, png)}

(PNG 11.4 MB)

Supplementary Material 3 (TIF 22.3 MB)^{(22.3MB, tif)}

Supplementary figure 16^{(375.2KB, png)}

(PNG 375 KB)

Supplementary Material 4 (TIF 1.65 MB)^{(16.1MB, tif)}

Supplementary figure 17^{(11.1MB, png)}

(PNG 11.0 MB)

Supplementary Material 5 (TIF 12.9 MB)^{(12.9MB, tif)}

Supplementary Material 6 (JPG 1.49 MB)^{(1.5MB, jpg)}

Supplementary figure 18^{(8.4MB, png)}

(PNG 8.35 MB)

Supplementary Material 7 (TIF 16.1 MB)^{(16.1MB, tif)}

Supplementary figure 19^{(4.7MB, png)}

(PNG 4.65 MB)

Supplementary Material 8 (TIF 9.60 MB)^{(9.6MB, tif)}

Supplementary figure 20^{(4.8MB, png)}

(PNG 4.75 MB)

Supplementary Material 9 (TIF 9.96 MB)^{(10MB, tif)}

Supplementary figure 21^{(2.4MB, png)}

(PNG 2.36 MB)

Supplementary Material 10 (TIF 4.96 MB)^{(5MB, tif)}

Supplementary figure 22^{(6.4MB, png)}

(PNG 6.39 MB)

Supplementary Material 11 (TIF 19.7 MB)^{(19.7MB, tif)}

Supplementary figure 23^{(388.6KB, png)}

(PNG 388 KB)

Supplementary Material 12 (TIF 7.51 MB)^{(7.5MB, tif)}

Supplementary figure 24^{(362.6KB, png)}

(PNG 362 KB)

Supplementary Material 13 (TIF 2.75 MB)^{(2.8MB, tif)}

Supplementary figure 25^{(2.9MB, png)}

(PNG 2.93 MB)

Supplementary Material 14 (TIF 9.00 MB)^{(9MB, tif)}

Supplementary figure 26^{(1.7MB, png)}

(PNG 1.67 MB)

Supplementary Material 15 (TIF 2.49 MB)^{(2.5MB, tif)}

Supplementary figure 27^{(1.3MB, png)}

(PNG 1.30 MB)

Supplementary Material 16 (TIF 9.64 MB)^{(9.6MB, tif)}

10753_2026_2498_MOESM17_ESM.xlsx^{(11.4KB, xlsx)}

Supplementary Material 17 (XLSX 11.4 KB)

10753_2026_2498_MOESM18_ESM.xlsx^{(11KB, xlsx)}

Supplementary Material 18 (XLSX 11.0 KB)

Supplementary Material 19 (PDF 6.24 MB)^{(6.2MB, pdf)}

Supplementary figure 28^{(2.2MB, png)}

(PNG 2.20 MB)

Supplementary Material 20 (TIF 4.58 MB)^{(4.6MB, tif)}

10753_2026_2498_MOESM21_ESM.xlsx^{(10.7KB, xlsx)}

Supplementary Material 21 (XLSX 10.7 KB)

Acknowledgements

Thank you to all the laboratory colleagues who provided support for this study.

Abbreviations

AQP1: Aquaporin 1
AAV: Adeno-associated virus
Arg1: Arginase 1
ASC: Apoptosis-associated speck-like protein containing a CARD
CSD: Chronic sleep deprivation
CD86: Cluster of differentiation 86
CD206: Cluster of differentiation 206
Curcumin: diferuloylmethane
cleaved Caspase1: Cleaved Caspase-1
GFAP: Glial fibrillary acidic protein
IBA1: Ionized calcium-binding adapter molecule 1
LPS: Lipopolysaccharide
MWM: Morris water maze
mRNA: Messenger RNA
NORT: Novel object recognition test
NeuN: Neuronal nuclei antigen
NC: Negative control
NLRP3: NOD-like receptor family pyrin domain containing 3
Nig: Nigericin
siRNA: Small interfering RNA
Y-maze: Y-maze test

Author Contributions

Yanhong Xiong, Jun Ying and Fuzhou Hua acquired and analyzed the data, drafted the figures and wrote the manuscript; Yanhong Xiong, Weidong Liang, Xifeng Wang, Hong Zhu, Guan Xilong and Pengcheng Yi discussed the results, contributed greatly to the research design and made critical modifications to the manuscript; Yanhong Xiong, Weidong Liang, Xifeng Wang, Hong Zhu, Guan Xilong, Pengcheng Yi, Lieliang Zhang, Yueyang You, and Yingchuan Hu conducted the experiments; Xifeng Wang and Hong Zhu assisted in the statistical analyses; Pengcheng Yi, Lieliang Zhang, and Xifeng Wang provided technical support; and Yanhong Xiong, Weidong Liang, Hong Zhu, Jun Ying and Fuzhou Hua designed the study and revised the manuscript. We want to acknowledge our team of researchers and clinicians for their strong support of this work. All the authors read and approved the final manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (82401426, 82571377, 82460231, and 82271234), the Natural Science Foundation of Jiangxi Province (20242BAB26133 and 20243BCE51077), the Jiangxi Province Key Laboratory of Anesthesiology (2024SSY06161), and the Science Foundation of Yingtan City (20244–390331).

Data Availability

The raw data that support the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethical Approval and Consent to Participate

The current study does not involve human participants, human tissue or human data. All animal experiments conducted in this research were approved by the Ethics Committee Animal Board (Approval number: RYE2024032901). We followed the institutional guidelines rigorously during the study.

Consent for Publication

All the authors have read and approved the final manuscript.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Yanhong Xiong and Weidong Liang contributed equally.

Contributor Information

Xifeng Wang, Email: ndyfy04789@ncu.edu.cn.

Fuzhou Hua, Email: huafuzhou@126.com.

References

1.Cheng, Z. et al. 2025. The niacin receptor HCAR2 prevents sleep deprivation-induced cognitive impairment by inhibiting microglia-mediated neuroinflammation in mice. International Immunopharmacology 162:115174. [DOI] [PubMed] [Google Scholar]
2.Becker, S. P. et al. 2018. Sleep in a large, multi-university sample of college students: sleep problem prevalence, sex differences, and mental health correlates. Sleep Health 4(2):174–181. [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Li, L. et al. 2017. Sleep Duration and Sleep Patterns in Chinese University Students: A Comprehensive Meta-Analysis. Journal Of Clinical Sleep Medicine 13(10):1153–1162. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Lee, S. W. H., K. Y. Ng, and W. K. Chin. 2017. The impact of sleep amount and sleep quality on glycemic control in type 2 diabetes: A systematic review and meta-analysis. Sleep Medicine Reviews 31:91–101. [DOI] [PubMed] [Google Scholar]
5.Li, X., et al. 2023. Sleep deprivation promotes endothelial inflammation and atherogenesis by reducing exosomal miR-182-5p. Arteriosclerosis, Thrombosis, and Vascular Biology 43 (6): 995–1014. [DOI] [PubMed] [Google Scholar]
6.Holth, J. K. et al. 2019. The sleep-wake cycle regulates brain interstitial fluid tau in mice and CSF tau in humans. Science 363(6429):880–884. [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Leong, R. L. F., and M. W. L. Chee. 2023. Understanding the Need for Sleep to Improve Cognition. Annu Rev Psychol 74:27–57. [DOI] [PubMed] [Google Scholar]
8.Zhai, Q. et al. 2023. Reducing complement activation during sleep deprivation yields cognitive improvement by dexmedetomidine. British Journal Of Anaesthesia 131(3):542–555. [DOI] [PubMed] [Google Scholar]
9.Jiang, N. et al. 2023. Tenuifolin ameliorates the sleep deprivation-induced cognitive deficits. Phytotherapy Research: Ptr 37(2):464–476. [DOI] [PubMed] [Google Scholar]
10.Guo, C. et al. 2024. EPO Deficiency Upregulates GADD45b/p38 MAPK Axis, Mediating Schizophrenia-Related Synaptic and Cognitive Impairments. Adv Sci (Weinh) 11(47):e2406979. [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Zhou, L., et al. 2024. N6-methyladenosine demethylase FTO regulates neuronal oxidative stress via YTHDC1-ATF3 axis in arsenic-induced cognitive dysfunction. Journal of Hazardous Materials 480: 135736. [DOI] [PubMed] [Google Scholar]
12.Chen, J. et al. 2025. A CoQ10 analog ameliorates cognitive impairment and early brain injury after subarachnoid hemorrhage by regulating ferroptosis and neuroinflammation. Redox Biology 84:103684. [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Mela, V. et al. 2025. Microbiota fasting-related changes ameliorate cognitive decline in obesity and boost ex vivo microglial function through the gut-brain axis. Gut 74(11):1828–1846. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Gao, Y. et al. 2025. Poliumoside alleviates microglia-mediated inflammation and blood-brain barrier disruption via modulating the polarization of microglia after ischemic stroke in mice. Phytomedicine 143:156881. [DOI] [PubMed] [Google Scholar]
15.Fu, S. et al. 2024. Müller Cells Harboring Exosomal lncRNA OGRU Modulate Microglia Polarization in Diabetic Retinopathy by Serving as miRNA Sponges. Diabetes 73(11):1919–1934. [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Zhang, S. et al. 2024. Treadmill exercise improves cerebral ischemia injury by regulating microglia polarization via downregulation of MMP12. Int Immunopharmacol, 142(Pt B): p. 113210. [DOI] [PubMed]
17.Liu, Y., et al. 2025. Shenqi wan inhibits cellular senescence to alleviate renal fibrosis by modulating the AQP1/TGF-β1/ITPR1 axis. Phytomedicine 147: 157201. [DOI] [PubMed] [Google Scholar]
18.Benga, G. 2012. The first discovered water channel protein, later called aquaporin 1: Molecular characteristics, functions and medical implications. Molecular Aspects of Medicine 33 (5–6): 518–534. [DOI] [PubMed] [Google Scholar]
19.Ma, Y. et al. 2025. Aquaporin–1 regulates microglial polarization and inflammatory response in traumatic brain injury. International Journal Of Molecular Medicine, 55(3). [DOI] [PMC free article] [PubMed]
20.Yu, B., et al. 2020. Silencing of aquaporin1 activates the Wnt signaling pathway to improve cognitive function in a mouse model of Alzheimer’s disease.Gene 755: 144904. [DOI] [PubMed] [Google Scholar]
21.Chou, V. et al. 2023. INPP5D regulates inflammasome activation in human microglia. Nature Communications 14(1):7552. [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Yan, Y.Q., et al. 2023. Parkin regulates microglial NLRP3 and represses neurodegeneration in Parkinson’s disease. Aging Cell 22 (6): e13834. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Chen, Y., et al. 2024. WTAP participates in neuronal damage by protein translation of NLRP3 in an m6A-YTHDF1-dependent manner after traumatic brain injury. International Journal of Surgery 110 (9): 5396–5408. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Huang, X., et al. 2023. TRIM45 aggravates microglia pyroptosis via Atg5/NLRP3 axis in septic encephalopathy. Journal of Neuroinflammation 20 (1): 284. [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Li, X. et al. 2024. Aloe-emodin alleviates cerebral ischemia-reperfusion injury by regulating microglial polarization and pyroptosis through inhibition of NLRP3 inflammasome activation. Phytomedicine 129:155578. [DOI] [PubMed] [Google Scholar]
26.Wang, L., et al. 2024. NLRP3 inflammasome inhibition after pilocarpine-induced status epilepticus attenuates chronic inflammation in epileptic mice. Journal of Inflammation Research 17: 6143–6158. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Zhang, Y., et al. 2023. ChemR23 activation attenuates cognitive impairment in chronic cerebral hypoperfusion by inhibiting NLRP3 inflammasome-induced neuronal pyroptosis. Cell Death & Disease 14 (11): 721. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Tiwari, S. K. et al. 2014. Curcumin-loaded nanoparticles potently induce adult neurogenesis and reverse cognitive deficits in Alzheimer’s disease model via canonical Wnt/β-catenin pathway. Acs Nano 8(1):76–103. [DOI] [PubMed] [Google Scholar]
29.Li, G. et al. 2025. Curcumin reverses cognitive deficits through promoting neurogenesis and synapse plasticity via the upregulation of PSD95 and BDNF in mice. Scientific Reports 15(1):1135. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Gu, J. et al. 2024. Erythrocyte membrane-coated nanocarriers modified by TGN for Alzheimer’s disease. Journal Of Controlled Release : Official Journal Of The Controlled Release Society 366:448–459. [DOI] [PubMed] [Google Scholar]
31.Jin, T. et al. 2022. Curcumin can improve Parkinson’s disease via activating BDNF/PI3k/Akt signaling pathways. Food And Chemical Toxicology 164:113091. [DOI] [PubMed] [Google Scholar]
32.Cao, X., and Y. Pu. 2025. Curcumin Regulates Microglia Polarization to Alleviate Ischemic Stroke by Targeting microRNA-205-5p/Kruppel-Like Factor 2 (KLF2)/Activating Transcription Factor 2 (ATF2) Axis. Chemical Biology & Drug Design 105(1):e70050. [DOI] [PubMed] [Google Scholar]
33.Li, J. et al. 2023. Lipocalin-2-mediated astrocyte pyroptosis promotes neuroinflammatory injury via NLRP3 inflammasome activation in cerebral ischemia/reperfusion injury. J Neuroinflammation 20(1):148. [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Zhang, K. et al. 2025. 53 promote oxidative stress, neuroinflammation and behavioral disorders via DDIT4-NF-κB signaling pathway. Redox Biology 86:103836. [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Yin, C., et al. 2024. Mechanism of Salvia miltiorrhiza Bunge extract to alleviate chronic sleep deprivation-induced cognitive dysfunction in rats. Phytomedicine 130: 155725. [DOI] [PubMed] [Google Scholar]
36.Bao, H. et al. 2023. GABA induced by sleep deprivation promotes the proliferation and migration of colon tumors through miR-223-3p endogenous pathway and exosome pathway. Journal Of Experimental & Clinical Cancer Research : Cr 42(1):344. [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Zamore, Z., and S.C. Veasey. 2022. Neural consequences of chronic sleep disruption. Trends in Neurosciences 45 (9): 678–691. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Naidoo, N. et al. 2014. Aging and sleep deprivation induce the unfolded protein response in the pancreas: implications for metabolism. Aging Cell 13(1):131–141. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Feng, X., et al. 2021. LncRNA 4344 promotes NLRP3-related neuroinflammation and cognitive impairment by targeting miR-138-5p. Brain, Behavior, and Immunity 98: 283–298. [DOI] [PubMed] [Google Scholar]
40.Leger, M. et al. 2013. Object recognition test in mice. Nature Protocols 8(12):2531–2537. [DOI] [PubMed] [Google Scholar]
41.Jing, G. et al. 2025. OTUD1 exacerbates sepsis-associated encephalopathy by promoting HK2 mitochondrial release to drive microglia pyroptosis. J Neuroinflammation 22(1):154. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Hua, F., et al. 2023. β-arrestin1 regulates astrocytic reactivity via Drp1-dependent mitochondrial fission: Implications in postoperative delirium. Journal of Neuroinflammation 20 (1): 113. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Li, Y. et al. 2017. Propane-2-sulfonic acid octadec-9-enyl-amide, a novel PPARα/γ dual agonist, protects against ischemia-induced brain damage in mice by inhibiting inflammatory responses. Brain, Behavior, And Immunity 66:289–301. [DOI] [PubMed] [Google Scholar]
44.Tao, W. et al. 2021. Magnolol attenuates depressive-like behaviors by polarizing microglia towards the M2 phenotype through the regulation of Nrf2/HO-1/NLRP3 signaling pathway. Phytomedicine 91:153692. [DOI] [PubMed] [Google Scholar]
45.Bi, W. et al. 2023. PSMC5 regulates microglial polarization and activation in LPS-induced cognitive deficits and motor impairments by interacting with TLR4. J Neuroinflammation 20(1):277. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Zhang, H. et al. 2025. Cathepsin L Aggravates Neuroinflammation via Promoting Microglia M1 Polarization and NLRP3 Activation After Spinal Cord Injury. Faseb j 39(9):e70561. [DOI] [PubMed] [Google Scholar]
47.Aalderink, G. et al. 2026. Pro-inflammatory response of human iPSC-derived intestinal epithelial monolayers towards microbial toxins LPS and nigericin. Archives Of Toxicology 100(2):695–708. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Huang, H. et al. 2024. Tanreqing injection inhibits dengue virus encephalitis by suppressing the activation of NLRP3 inflammasome. Chinese Medicine 19(1):24. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Lan, Z. et al. 2024. Curcumin-primed olfactory mucosa-derived mesenchymal stem cells mitigate cerebral ischemia/reperfusion injury-induced neuronal PANoptosis by modulating microglial polarization. Phytomedicine 129:155635. [DOI] [PubMed] [Google Scholar]
50.Li, Q. et al. 2023. Impaired lipophagy induced-microglial lipid droplets accumulation contributes to the buildup of TREM1 in diabetes-associated cognitive impairment. Autophagy 19(10):2639–2656. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Choi, J. W. et al. 2024. Aerobic exercise attenuates LPS-induced cognitive dysfunction by reducing oxidative stress, glial activation, and neuroinflammation. Redox Biology 71:103101. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Liu, L. et al. 2024. Running exercise alleviates hippocampal neuroinflammation and shifts the balance of microglial M1/M2 polarization through adiponectin/AdipoR1 pathway activation in mice exposed to chronic unpredictable stress. Molecular Psychiatry 29(7):2031–2042. [DOI] [PubMed] [Google Scholar]
53.Leng, F., and P. Edison. 2021. Neuroinflammation and microglial activation in Alzheimer disease: Where do we go from here? Nature Reviews Neurology 17 (3): 157–172. [DOI] [PubMed] [Google Scholar]
54.Wang, Z. et al. 2021. Gut microbiota modulates the inflammatory response and cognitive impairment induced by sleep deprivation. Molecular Psychiatry 26(11):6277–6292. [DOI] [PubMed] [Google Scholar]
55.Yu, H. et al. 2023. Metabolic reprogramming and polarization of microglia in Parkinson’s disease: Role of inflammasome and iron. Ageing Research Reviews 90:102032. [DOI] [PubMed] [Google Scholar]
56.Parhizkar, S. et al. 2023. Sleep deprivation exacerbates microglial reactivity and Aβ deposition in a TREM2-dependent manner in mice. Science Translational Medicine 15(693):eade6285. [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Mukherjee, U. et al. 2024. Mechanisms, consequences and role of interventions for sleep deprivation: Focus on mild cognitive impairment and Alzheimer’s disease in elderly. Ageing Research Reviews 100:102457. [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Garbarino, S. et al. 2020. Circadian Rhythms, Sleep, Immunity, and Fragility in the Elderly: The Model of the Susceptibility to Infections. Frontiers In Neurology 11:558417. [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Li, N. et al. 2023. Akkermansia muciniphila supplementation prevents cognitive impairment in sleep-deprived mice by modulating microglial engulfment of synapses. Gut Microbes 15(2):2252764. [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Uddin, M. S. et al. 2020. Circadian and sleep dysfunction in Alzheimer’s disease. Ageing Research Reviews 60:101046. [DOI] [PubMed] [Google Scholar]
61.Wang, X. et al. 2023. Gut microbiota-derived metabolites mediate the neuroprotective effect of melatonin in cognitive impairment induced by sleep deprivation. Microbiome 11(1):17. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Long, Y. et al. 2024. Modulating the polarization phenotype of microglia - A valuable strategy for central nervous system diseases. Ageing Research Reviews 93:102160. [DOI] [PubMed] [Google Scholar]
63.Gao, M. et al. 2024. Targeted mRNA Nanoparticles Ameliorate Blood-Brain Barrier Disruption Postischemic Stroke by Modulating Microglia Polarization. Acs Nano 18(4):3260–3275. [DOI] [PubMed] [Google Scholar]
64.Zou, G. et al. 2025. Signal-induced NLRP3 phase separation initiates inflammasome activation. Cell Research 35(6):437–452. [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Xu, J. et al. 2025. NEK7 phosphorylation amplifies NLRP3 inflammasome activation downstream of potassium efflux and gasdermin D. Sci Immunol 10(103):eadl2993. [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Zeng, B. et al. 2022. Dextran sodium sulfate potentiates NLRP3 inflammasome activation by modulating the KCa3.1 potassium channel in a mouse model of colitis. Cellular & Molecular Immunology 19(8):925–943. [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Song, Z. et al. 2025. Innate immune sensing of Z-nucleic acids by ZBP1-RIPK1 axis drives neuroinflammation in Alzheimer’s disease. Immunity 58(10):2574–2592e9. [DOI] [PubMed] [Google Scholar]
68.Scheiblich, H. et al. 2024. Microglia rescue neurons from aggregate-induced neuronal dysfunction and death through tunneling nanotubes. Neuron 112(18):3106–3125e8. [DOI] [PubMed] [Google Scholar]
69.Yang, Y. et al. 2025. H4K12 lactylation potentiates mitochondrial oxidative stress via the Foxo1 pathway in diabetes-induced cognitive impairment. Journal Of Advanced Research 78:391–407. [DOI] [PMC free article] [PubMed] [Google Scholar]
70.D’Acierno, M., R. A. Fenton, and E. J. Hoorn. 2025. The biology of water homeostasis. Nephrology, Dialysis, Transplantation 40(4):632–640. [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Zhang, H. et al. 2026. Aquaporin 4 and its isoforms regulation ameliorate AQP4 Mis-localization-induced glymphatic dysfunction in ischemic stroke. Journal Of Advanced Research 80:835–849. [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Zhu, D. D. et al. 2023. AQP4 Aggravates Cognitive Impairment in Sepsis-Associated Encephalopathy through Inhibiting Na(v) 1.6-Mediated Astrocyte Autophagy. Adv Sci (Weinh) 10(14):e2205862. [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Feng, S. et al. 2023. High-intensity interval training ameliorates Alzheimer’s disease-like pathology by regulating astrocyte phenotype-associated AQP4 polarization. Theranostics 13(10):3434–3450. [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Villarreal, A. M. et al. 2025. Aquaporins in Allergic Response: A New Player. Clinical Reviews In Allergy And Immunology 68(1):71. [DOI] [PubMed] [Google Scholar]
75.Zhao, Y. et al. 2024. AQP1 Deficiency Drives Phthalate-Induced Epithelial Barrier Disruption through Intestinal Inflammation. Journal Of Agriculture And Food Chemistry 72(27):15334–15344. [DOI] [PubMed] [Google Scholar]
76.Liu, Y. et al. 2022. Brain-Targeted Biomimetic Nanodecoys with Neuroprotective Effects for Precise Therapy of Parkinson’s Disease. ACS Cent Sci 8(9):1336–1349. [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Nazir, F. H. et al. 2018. Expression and secretion of synaptic proteins during stem cell differentiation to cortical neurons. Neurochemistry International 121:38–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
78.Yang, P. et al. 2024. Precise Modulation of Pericyte Dysfunction by a Multifunctional Nanoprodrug to Ameliorate Alzheimer’s Disease. Acs Nano 18(22):14348–14366. [DOI] [PubMed] [Google Scholar]
79.Wang, L., et al. 2024. Curcumin-loaded polysaccharide microparticles alleviated DSS-induced ulcerative colitis by improving intestinal microecology and regulating MAPK/NF-κB/Nrf2/NLRP3 pathways. International Journal of Biological Macromolecules 281 (Pt 4): 136687. [DOI] [PubMed] [Google Scholar]
80.Xie, Y. L. et al. 2017. Curcumin attenuates lipopolysaccharide/d-galactosamine-induced acute liver injury by activating Nrf2 nuclear translocation and inhibiting NF-kB activation. Biomedicine & Pharmacotherapy 91:70–77. [DOI] [PubMed] [Google Scholar]
81.Dai, C. et al. 2023. Involvement of the inhibition of mitochondrial apoptotic, p53, NF-κB pathways and the activation of Nrf2/HO-1 pathway in the protective effects of curcumin against copper sulfate-induced nephrotoxicity in mice. Ecotoxicology And Environmental Safety 249:114480. [DOI] [PubMed] [Google Scholar]